Bid-Ask Imbalance: Volume-Weighted Spread Changes as Trading Signals

Decoding Market Signals — How Bid-Ask Imbalance Predicts Trading Opportunities

Quantitative trading success depends on signals hidden within order book dynamics. One of the most insightful is bid-ask imbalance — the difference between the highest price buyers are willing to pay (bid) and the lowest price sellers will accept (ask), weighted by trade volume.

Understanding the Power of Bid-Ask Imbalance

The bid-ask spread is a constant negotiation between buyers and sellers. Monitoring how it changes — especially in conjunction with traded volume — provides early information about market liquidity, volatility, and where price is heading next.

Three patterns repeat across instruments and time:

Building the Feature: Volume-Weighted Bid-Ask Spread Change

Our feature bid_ask_spread_change_volume_weighted computes spread changes weighted by trade volumes at each timestamp, using the top 5 levels of the order book on each side.

Step 1: Computing Weighted Prices

We compute volume-weighted bid and ask prices from the top 5 levels on each side of the order book. Averaging the price-volume pairs gives a stable picture of where market depth actually sits.

Step 2: Calculating Spread and Mid Price

Using the weighted prices we derive two scalars per timestamp.

Step 3: Integrating Trade Volumes

We integrate live trade flow with the weighted spread. Each trade's quantity multiplied by its price deviation from the mid gives a per-trade contribution to the volume-weighted spread change. Summing these contributions across the timestamp produces the final per-bar metric — capturing both how the spread is moving and how much volume is moving with it.

Backtest Performance

We backtested the feature inside our ML trading system on BTC spot, with realistic execution conditions (round-trip transaction cost of 0.004% and slippage of 0.0002%). Across the test window:

Practical Insights — From Raw Data to Trading Signals

Three practical use cases emerged from the backtest review:

Conclusion: Harnessing Bid-Ask Imbalance for Strategic Advantage

The bid-ask imbalance feature is not a stand-alone trading system — it is one signal in a stack. What makes it valuable is that it surfaces information before price has fully moved: institutional footprints show up in the order book first, in trade-volume distribution second, and in mid-price last. A signal that triangulates the first two has a small but persistent timing edge over one that only watches the third.

Key Takeaways

• Volume-weighted signals detect institutional footprints, revealing what "smart money" is doing before price movements become obvious.

• Asymmetric spread changes offer early warning of directional moves — narrow spreads with high volume often precede strong moves.

• Systematic implementation removes emotional bias from trading decisions, focusing on order book mathematics.

• Differential performance across regimes provides input for adaptive trading systems.

Next Steps and Further Research

In upcoming posts we will cover:

• Combining this feature with complementary signals (momentum, regime classification, news velocity).

• Optimizing parameters per asset class and timeframe.

• Applying reinforcement learning to dynamically adjust trading on the signal.